MeiliSearch/benchmarks/Cargo.toml

[package]
name = "benchmarks"
publish = false

version.workspace = true
authors.workspace = true
description.workspace = true
homepage.workspace = true
readme.workspace = true
edition.workspace = true
license.workspace = true

[dependencies]
anyhow = "1.0.79"
csv = "1.3.0"
milli = { path = "../milli" }
mimalloc = { version = "0.1.39", default-features = false }
serde_json = { version = "1.0.111", features = ["preserve_order"] }

[dev-dependencies]
criterion = { version = "0.5.1", features = ["html_reports"] }
rand = "0.8.5"
rand_chacha = "0.3.1"
roaring = "0.10.2"

[build-dependencies]
anyhow = "1.0.79"
bytes = "1.5.0"
convert_case = "0.6.0"
flate2 = "1.0.28"
reqwest = { version = "0.11.23", features = ["blocking", "rustls-tls"], default-features = false }

[features]
default = ["milli/all-tokenizations"]

[[bench]]
name = "search_songs"
harness = false

[[bench]]
name = "search_wiki"
harness = false

[[bench]]
name = "search_geo"
harness = false

[[bench]]
name = "indexing"
harness = false
move the benchmarks to another crate so we can download the datasets automatically without adding overhead to the build of milli 2021-05-25 17:09:14 +02:00			`[package]`
			`name = "benchmarks"`
			`publish = false`

Use the workspace inheritance feature of rust 1.64 2023-02-15 13:51:07 +01:00			`version.workspace = true`
			`authors.workspace = true`
			`description.workspace = true`
			`homepage.workspace = true`
			`readme.workspace = true`
			`edition.workspace = true`
			`license.workspace = true`

move the benchmarks to another crate so we can download the datasets automatically without adding overhead to the build of milli 2021-05-25 17:09:14 +02:00			`[dependencies]`
Upgrade all compatible dependencies 2024-01-16 15:05:03 +01:00			`anyhow = "1.0.79"`
			`csv = "1.3.0"`
Allow to disable specialized tokenizations (again) In PR #2773, I added the `chinese`, `hebrew`, `japanese` and `thai` feature flags to allow melisearch to be built without huge specialed tokenizations that took up 90% of the melisearch binary size. Unfortunately, due to some recent changes, this doesn't work anymore. The problem lies in excessive use of the `default` feature flag, which infects the dependency graph. Instead of adding `default-features = false` here and there, it's easier and more future-proof to not declare `default` in `milli` and `meilisearch-types`. I've renamed it to `all-tokenizers`, which also makes it a bit clearer what it's about. 2023-04-24 00:26:08 +02:00			`milli = { path = "../milli" }`
Upgrade all compatible dependencies 2024-01-16 15:05:03 +01:00			`mimalloc = { version = "0.1.39", default-features = false }`
			`serde_json = { version = "1.0.111", features = ["preserve_order"] }`
Use mimalloc for benchmarks on macOS 2022-08-10 12:31:09 +02:00
move the benchmarks to another crate so we can download the datasets automatically without adding overhead to the build of milli 2021-05-25 17:09:14 +02:00			`[dev-dependencies]`
Update criterion to 0.5.1 to remove the atty dependency 2023-07-03 17:09:29 +02:00			`criterion = { version = "0.5.1", features = ["html_reports"] }`
Add deletion benchmarks 2022-06-16 10:17:58 +02:00			`rand = "0.8.5"`
			`rand_chacha = "0.3.1"`
Upgrade all compatible dependencies 2024-01-16 15:05:03 +01:00			`roaring = "0.10.2"`
move the benchmarks to another crate so we can download the datasets automatically without adding overhead to the build of milli 2021-05-25 17:09:14 +02:00
			`[build-dependencies]`
Upgrade all compatible dependencies 2024-01-16 15:05:03 +01:00			`anyhow = "1.0.79"`
			`bytes = "1.5.0"`
Upgrade all dependencies 2022-10-04 11:29:39 +02:00			`convert_case = "0.6.0"`
Upgrade all compatible dependencies 2024-01-16 15:05:03 +01:00			`flate2 = "1.0.28"`
			`reqwest = { version = "0.11.23", features = ["blocking", "rustls-tls"], default-features = false }`
move the benchmarks to another crate so we can download the datasets automatically without adding overhead to the build of milli 2021-05-25 17:09:14 +02:00
Make milli's default features optional in other executable targets 2022-08-16 20:02:46 +02:00			`[features]`
Allow to disable specialized tokenizations (again) In PR #2773, I added the `chinese`, `hebrew`, `japanese` and `thai` feature flags to allow melisearch to be built without huge specialed tokenizations that took up 90% of the melisearch binary size. Unfortunately, due to some recent changes, this doesn't work anymore. The problem lies in excessive use of the `default` feature flag, which infects the dependency graph. Instead of adding `default-features = false` here and there, it's easier and more future-proof to not declare `default` in `milli` and `meilisearch-types`. I've renamed it to `all-tokenizers`, which also makes it a bit clearer what it's about. 2023-04-24 00:26:08 +02:00			`default = ["milli/all-tokenizations"]`
Make milli's default features optional in other executable targets 2022-08-16 20:02:46 +02:00
move the benchmarks to another crate so we can download the datasets automatically without adding overhead to the build of milli 2021-05-25 17:09:14 +02:00			`[[bench]]`
prefix the search benchmarks with 'search' 2021-07-29 14:30:33 +02:00			`name = "search_songs"`
move the benchmarks to another crate so we can download the datasets automatically without adding overhead to the build of milli 2021-05-25 17:09:14 +02:00			`harness = false`

			`[[bench]]`
prefix the search benchmarks with 'search' 2021-07-29 14:30:33 +02:00			`name = "search_wiki"`
move the benchmarks to another crate so we can download the datasets automatically without adding overhead to the build of milli 2021-05-25 17:09:14 +02:00			`harness = false`
add benchmarks for indexing 2021-07-07 11:42:14 +02:00
add benchmarks for the geosearch 2021-09-13 18:08:28 +02:00			`[[bench]]`
			`name = "search_geo"`
			`harness = false`

add benchmarks for indexing 2021-07-07 11:42:14 +02:00			`[[bench]]`
			`name = "indexing"`
			`harness = false`